Real-Time Mapping and Tree Measurements Using UAVs

Precise tree measurement is essential for forest inventory and biomass estimation. Current methods often capture data only from the upper or lower parts of trees, resulting in incomplete or estimated measurement, bringing uncertainty to measurements. Some approaches fuse upper and lower tree data, but they require the alignment and merging of dense point clouds, making large-scale implementation challenging. This study presents an autonomous navigation and real-time data extraction method using an unmanned aerial vehicle (UAV) equipped with depth cameras and a LiDAR sensor. The system navigates autonomously and maps the environment around it, using a uniform voxel grid to segment and measure individual trees. The results demonstrated that: (1) the UAV successfully navigates autonomously between trees, mapping the unstructured and unknown environment while performing realtime reconstruction; (2) tree trunk segmentation, measurement, and localization were achieved with an root mean square error (RMSE) of 0.22 m; and (3) tree height measurements obtained an RMSE of 0.05 m. The proposed methodology proved to be effective for forest inventory applications, providing accurate tree measurements with improved computational efficiency.